Not Enough Data to Be Fair? Evaluating Fairness Implications of Data Scarcity Solutions
Journal
Proceedings of the 58th Hawaii International Conference on System Sciences (HICSS)
Type
conference paper
Date Issued
2025-01-07
Author(s)
Research Team
IWI6
Abstract
This study explores the implications of the use of data scarcity solutions on fairness in machine learning, specifically in consumer credit interest rate prediction. We develop a comprehensive taxonomy of Data Scarcity Solutions (DSS) by analyzing academic literature, data science competitions, and practical implementations. We identify six distinct DSS clusters: Data Extension, Pre-Training, Public Data Inclusion, Data Sharing, Federated Learning, and Active Learning. Our evaluation shows that most DSS enhance both performance and fairness, with minimal negative correlation between the two. Notably, approaches incorporating external or synthetic data significantly improve fairness. This research contributes to understanding DSS beyond algorithmic performance, providing a framework for evaluating their societal impact. Furthermore, it offers practitioners a taxonomy to select the right method for tackling data scarcity and addresses fairness concerns in real-world scenarios.
Language
English
Keywords
Data Scarcity
Fairness
Machine Learning
Consumer Credit
Taxonomy
HSG Classification
contribution to scientific community
Refereed
Yes
Publisher place
Hawaii, USA
Start page
6886
End page
6895
Pages
10
Event Title
Proceedings of the 58th Hawaii International Conference on System Sciences (HICSS)
Event Location
Hawaii, USA
Event Date
07.01.2025
Subject(s)
Division(s)
File(s)![Thumbnail Image]()
open.access
Name
JML_1014.pdf
Size
691.4 KB
Format
Adobe PDF
Checksum (MD5)
2c3f3c02d165563185fa878612822ac3